山西方言与普通话进行司法话者识别可用特征音段研究

发布时间：2019-03-24 17:10

【摘要】：近年来,利用人的语音来识别其身份的司法话者识别技术越来越成为打击犯罪、保护公民权益的一种重要技术手段。但随着犯罪人反侦查能力的提高或被告人趋利避害心理的影响,他们在违法犯罪或侵害他人合法权益时使用方言或普通话,而在收集比对样本时,经常会因这些人的不配合而得不到其对应的方言或普通话语音样本,只能在普通话与方言之间进行比对。截至目前,话者识别专家可用的特征音段首先是局限在同一语段当中的相同音段(语音相同的语句),其次是声韵调均相同的孤立的词组或句末音节、句首音节。由于特殊地理及历史等因素的影响,山西方言与普通话具有比官话与普通话更大的差异性。在山西方言与普通话之间进行司法话者识别时,语音检材与样本中很难找到满足现有条件的足够特征音段来进行比对。在这种情况下发掘其它可供比对的特征音段,则成为解决这一问题的关键所在。本文采用文献资料比对、田野调查、实验语音学分析及统计分析等方法,对话者自动识别在山西方言与普通话间进行司法话者识别的可行性,山西方言与普通话语音系统的差异性,山西汾城方言紧喉音节与对应普通话音节在司法话者识别时的可比性,同声同韵异调音节、异声同韵音节及不同后响复韵母共有响元音进行司法话者识别的可行性,利用非语音信息辅助司法话者识别等进行了较详细研究,得出了以下结论：(1)现阶段,企图只靠话者自动识别系统进行司法话者识别的想法不现实,而在山西方言与普通话之间进行司法话者识别更是难以达到令人满意的效果。在山西方言与普通话之间进行司法话者识别主要还要依靠专家方法。(2)山西方言中,与普通话韵母一致度最低的方言点分别为：中区的交城(5个),北区的平鲁(6个),东北区的广灵(12个),东南区的阳城(6个),南区的万荣(11个),西区的静乐(6个)、中阳(6个),进一步探究山西方言与普通话进行说话人识别的判阈应从中区的交城,北区的平鲁,东南区的阳城,西区的静乐、中阳这几个方言点着手。(3)在山西方言与普通话之间进行司法话者识别时,可利用的语音特征音段的选用顺序为：同声同韵同调音节→同声同韵异调音节→发声部位相同、方法不同的异声同韵同调音节→发声部位相同、方法不同的异声同韵异调音节→发声部位相近、方法相同的异声同韵同调音节→发声部位相近、方法相同的异声同韵异调音节→发声部位较远、方法相同的异声同韵同调音节→同调不同后响复韵母共有响元音→异调不同后响复韵母共有响元音。但在选用这些特征音段时还需要注意一些事项。(4)在方言与普通话之间进行司法话者识别时,可以充分利用非语音信息所揭示出的个体特性辅助解决话者识别问题。这些结论对进一步完善汉语的话者识别理论、解决办案过程中所遇到的实际问题具有重要价值,对方言研究与实际应用相结合,更好地发挥方言研究成果的社会效益,具有重要意义,对其他方言与普通话间的说话人识别研究,也具有重要的借鉴意义。
[Abstract]:In recent years, the use of people's voice to identify the judicial speech recognition technology of its identity has become an important technical means to fight crime and protect the rights and interests of citizens. However, with the improvement of the anti-investigation ability of the offender or the influence of the defendant's psychological influence, they use the dialect or the Putonghua in the case of the crime or the infringement of the legitimate rights and interests of others, and when the sample is collected, It is often not possible to get the corresponding dialect or the Putonghua voice sample due to the non-cooperation of these people, and can only be compared between the Putonghua and the dialect. At present, the speech segment of the speech recognition expert can first be the same sound segment (the same sentence as the voice) which is limited in the same segment, and the second is the same isolated phrase or the end syllable and the first syllable in the same language segment. Due to the influence of special geography and history, Shanxi dialect and Mandarin have a greater difference than Mandarin and Mandarin. In the case of the identification of the judicial speech between the Shanxi dialect and the Putonghua, it is difficult to find enough features to meet the existing conditions to make the comparison. In this case, it is the key to solve this problem. In this paper, the author uses the methods of literature comparison, field investigation, experimental phonetics analysis and statistical analysis to identify the feasibility, the difference between the Shanxi dialect and the Putonghua speech system, The comparison between the syllable of the compact throat of the Fencheng dialect of Shanxi and the corresponding Putonghua syllable in the identification of the legal person is the feasibility of the identification of the judicial speaker in the same rhyme, the same rhyme and the different after-sound and the complex finals. The following conclusions are drawn: (1) At present, it is not realistic to attempt to identify the system by the speaker automatically. However, it is difficult to achieve satisfactory effect in that identification of the judicial speech between the dialect of Shanxi and the Putonghua. The identification of the judicial speech between the dialect of Shanxi and the Putonghua mainly depends on the expert method. (2) In the Shanxi dialect, the dialect points with the lowest degree of conformity with the Chinese finals are: the Jiaocheng (5) in the Central District, the Pinglu (6) in the North District, the Guangling (12) in the northeast, the Yangcheng (6) in the south-east, the Wanrong (11 in the Southern District) and the Jingle in the west (6). Zhongyang (6), further explore the judgment threshold of the speaker recognition in Shanxi dialect and Putonghua from the intersection of the Central District, the Pinglu of the North District, the Yangcheng of the southeast, the Jingle in the Western District, and the dialect points of Zhongyang. (3) In the case of the identification of the judicial speech between the dialect of Shanxi and the Putonghua, the selection order of the sound segment of the speech feature that can be used is as follows: the same tone and the same tone of the same rhyme are the same as the same, and the different sounds of the same tone are the same as that of the same tone and the same tone. The method of the invention has the advantages that the abnormal sound of different sounds is similar to the sound-producing part of the same-tone and the same-tone, the same method is similar to that of the same-tone, In the same way, the same tone of the same tone is the same as that of the same tone, and the same tone is the same as that of the same tone. However, attention needs to be paid to the selection of these features. And (4) when the dialect and the Putonghua are recognized, the problem of the speaker identification can be assisted by taking full advantage of the individual characteristics disclosed by the non-voice information. These conclusions are of great significance to further improve the recognition theory of the Chinese language, solve the practical problems encountered in the process of handling the case, and combine the study of the dialects with the practical application to better play the social and social benefits of the research results of the dialects. It is also of great significance to identify and study the speaker between other dialects and Putonghua.
【学位授予单位】：山西大学
【学位级别】：博士
【学位授予年份】：2015
【分类号】：H116

【相似文献】